PIC18F CCP (PWM) and Timers1/3 interrupts problem

Alel8 · Joined: 20 Jan 2017 Posts: 2

Hello everyone,

PCM programmer and Ttelmah, THANKS because you have solved many problems I had, and by searching on the forum you already solved the problem. But I need your help now, please!

I have a PIC18F2685, and I'm using the CCP1 to produce a PWM signal. I'm also using timer_0 as a millisecond event counter.

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

Can you post a complete test program that shows the problem ?
In other words, if the problem is that enabling Timer3 shuts down CCP1,
then post a program that demonstrates this. Post the #include for the
PIC, #fuses, #use delay(), main(), etc. It should be compilable.
Leave out any code that isn't related to the problem.

Also post your compiler version.

Ttelmah · Joined: 11 Mar 2010 Posts: 19520

It certainly shouldn't affect the CCP.

The classic thing is to simplify. Just generate a minimum program generating the PWM, and running Timer3. Does it work?. If not, then this becomes something you can post so that we can look at the problem.

The obvious thing that changes, is whatever is in '//Do Things', so this would be the first place to look. If you remark out this code, what happens?. Obvious things to look for here:
Anything changing TRIS on the PWM port.
Anything touching the timers.
Anything accessing arrays of using pointers (which might 'overflow' and therefore affect other registers.

Where are you setting up the PWM?. Before or after you setup Timer3?.
The other things that may be happening, is that the code is simply never reaching the PWM setup. If you are executing too many interrupts, and they contain code that takes significant time, it's possible that effectively the main code is never actually getting executed. You mention a 1mSec counter, and show this using the RTC as an /256 counter. This implies potentially an interrupt every 256 instructions. If this is executing a handler, even if this does nothing at all, immediately 25% of your processor's time will be being used handling this. If you then add another interrupt at 45mSec, and this takes perhaps 35mSec to execute it's code, suddenly there is no time let to do anything.
It takes quite a bit of time to get into and out of an interrupt handler. Typically something like 60 instructions. This is why it is better to keep the number of interrupts low if possible. If you want a timer, then consider if you could do this by just reading the timer registers, rather than having an interrupt. If you have a 1mSec timer, and want a 45mSec one, then just add a counter in the 1mSec interrupt (uses less time provided it is a simple int8, than calling another interrupt). However many interrupts you have though repeat the mantra fifty times _spend as little time as possible inside the handler_.

As one further comment to this, if 256 instructions is about 1mSec, this implies the chip is only being clocked at 1MHz. Honestly if you want to have an interrupt at 1mSec, you need to be clocking the chip faster than this. However your comment on Timer3 setup, says you are using 16MHz, which would imply timer1 interrupting every 64uSec...

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

He said in a PM to me that upon trying to make a test progarm, it all
started working fine.

Alel8 · Joined: 20 Jan 2017 Posts: 2

Ttelmah · Joined: 11 Mar 2010 Posts: 19520

Your timer0 code is never going to be very accurate. You are attempting to tweak the loaded value to adjust for the instructions it takes to go into the interrupt (fair enough)_, but if another interrupt occurs, the tweak will be wrong.
Why do this?. You already have Timer2 ticking at exactly 1000HZ. Pointless to setup another timer and try to 'fiddle' it's timings. Just put the timer2 postscaler down to 1, and have a timer2 interrupt like: